The Big DevOps Misunderstanding

submited by

Style Pass

2023-01-24 12:00:07

At that time, IaaS and especially PaaS matured and allowed developers to deploy their applications themselves. So they no longer had to “throw” their application code over the fence to the operations department. Instead, software engineers could deploy their applications to production themselves.

Somewhere along the way, this idea was misunderstood to the point that the wrong definition of DevOps became the right one. We now have specific roles or departments that write “infrastructure as code” or “pipeline as code” to deploy applications.

This is not DevOps, but a further development of Systems Operations (SysOps). Why? Because writing the application and deploying it is still done by different roles or departments.

This trend is reinforced by Kubernetes and the fact that everyone thinks they need the same architecture as Netflix, Twitter, … In the end, people end up with a very complex setup to handle the load a simple $50 hosting package from the 90s could have done.

If you are a small company, try to find a platform (Heroku, Render, …) your developers can use to deploy their application themselves. This means they should not actually deploy them manually but they should write a pipeline to do so. If the chosen platform is good, it comes with a CD solution already. If not, you (as a developer) write the pipeline yourself. It’s no rocket since. If you can write Java code, you can as well copy & paste a few lines of yaml together. Of course you will also have to come up with a branching strategy and make sure the chosen platform is performing a backup of the databases. Application performance monitoring is a topic as well. But once configured, those things should not cause much of a trouble.